FR: Make a data frame from a (possibly named) vector or list

[jennybc]

Here's something I do fairly often, mostly with a list, but sometimes with a vector: Initialize a data frame with that list or vector as a variable and, at the same time, promote its names to a proper variable. Or, perhaps, add a variable of row numbers. Why is it so important to add the names or row numbers? Because later you'll want to process with tidyr, i.e. with unnest() and/or spread().

I could point to some real uses if I need to really sell this. But hopefully this will just make sense. Or someone will tell me it's already easy to do? It is already easy, but perhaps worth making a function for.

library(tibble)

x <- list(alpha = 'horrible', beta = 'list', gamma = 'column')

## wish it were easy to make the names a proper variable
data_frame(id = names(x), thing = x)
#> Source: local data frame [3 x 2]
#> 
#>      id    thing
#>   (chr)   (list)
#> 1 alpha <chr[1]>
#> 2  beta <chr[1]>
#> 3 gamma <chr[1]>

## where id can easily default to row number
data_frame(id = seq_along(x), thing = x)
#> Source: local data frame [3 x 2]
#> 
#>      id    thing
#>   (int)   (list)
#> 1     1 <chr[1]>
#> 2     2 <chr[1]>
#> 3     3 <chr[1]>

[krlmlr]

What verb would you use for this operation?

How about:

x %>% as_data_frame %>% tidyr::gather(id, thing)

Need to coerce to list if x is a vector. The following doesn't work:

x %>% tidyr::gather(id, thing)

I think this could be fixed by implementing gather_.list <- function(x, ...) gather_(tibble::data_frame(x), ...).

EDIT: The above also doesn't work if x is unnamed, but here you could use x %>% data_frame(thing=.) %>% add_rownames.

EDIT²: I think a new verb would help here, I'm not sure if this belongs here or in tidyr.

[jennybc]

I will try to propose a name.

[jennybc]

The more I think about it, maybe it makes sense to think about this as a treatment applied to a variable during the construction of a data frame. A way to say "add this variable AND promote its names to a proper variable". And also give some nice way of getting row numbers into the data frame? I have found dplyr::row_number() to be quite confusing / disappointing.

I realize id() is probably already overloaded with meaning already. upname() isn't great either, but hopefully this conveys the idea.

library(tibble)
x <- list(alpha = 'horrible', beta = 'list', gamma = 'column')

What if something like this:

df <- data_frame(id(x, "greek"))
## or
df <- data_frame(upname(x, "greek"))

produced this result

data_frame(greek = names(x), x = x)
#> Source: local data frame [3 x 2]
#> 
#>   greek        x
#>   (chr)   (list)
#> 1 alpha <chr[1]>
#> 2  beta <chr[1]>
#> 3 gamma <chr[1]>

I also wish it were easier to get plain row numbers. I wish this is what row_number() did but clearly it does not. So, again, fiction! What if something like this:

df <- data_frame(i = row_number(), upname(x, "greek"))

produced something like this:

data_frame(i = seq_along(x),
           greek = names(x),
           x = x)
#> Source: local data frame [3 x 3]
#> 
#>       i greek        x
#>   (int) (chr)   (list)
#> 1     1 alpha <chr[1]>
#> 2     2  beta <chr[1]>
#> 3     3 gamma <chr[1]>

[jennybc]

In addition to id() and upname(), dub() is a possible name for this name-promoting variable-pre-processing function.

[krlmlr]

Row numbers: Have you seen #11? Your example would be then

data_frame(...) %>% rownames_to_column("i")

How would you like:

dub <- function(x) as_data_frame(setNames(list(names(x), x), c("name", "value")))

This allows at least the creation of a two-column data frame from a named object, which then can be massaged further with the other dplyr verbs, and combined with other data frames using cbind().

@hadley: Would this perhaps be suitable for purrr:

unzip_names <- function(x) set_names(list(x, names(x)), c("name", "value"))
zip_names <- function(x) set_names(x[[1]], x[[2]])

[jennybc]

Sorry I can't really tell what #11 does just from reading the discussion. But I take your word for it that it would add the integers 1 through nrows(.) as variable i. Would it add as the first or last variable? Feels like you usually want it at the very front.

library(tibble)
x <- list(alpha = 'horrible', beta = 'list', gamma = 'column')
dub <- function(x) as_data_frame(setNames(list(names(x), x), c("name", "value")))
dub(x)
#> Source: local data frame [3 x 2]
#> 
#>    name    value
#>   (chr)   (list)
#> 1 alpha <chr[1]>
#> 2  beta <chr[1]>
#> 3 gamma <chr[1]>

The variables themselves and the object look great. Would dub() gain some arguments or different defaults in order to produce less generic names?

UPDATE: I think x and names of x are reversed in those speculative purrr functions.

[hadley]

What if this was just the as_data_frame() method for vectors?

[krlmlr]

@jennybc: purrr: Right, revised definition below.

unzip_names <- function(x) set_names(list(names(x), x), c("name", "value"))

rownames_to_column() will add to the front. Actually, we already have add_rownames() in dplyr, but it does "too much" and will be deprecated in favor of the new functions.

Defaults: I think we should support them, even if the renaming could be handle with a simple rename() step. tidyr::gather() has them too.

@hadley: Is there a dispatch for vector, as in as_data_frame.vector()? Otherwise it looks like we need to have as_data_frame.logical(), as_data_frame.character(), ...; we also need as_data_frame.Date(), as_data_frame.POSIXt(), ..., all with the same implementation. It also includes a certain amount of surprise -- if functions use as_data_frame() to convert input, the column names are auto-generated and the user might be unaware of it. Do you think it's worth it?

[hadley]

@krlmlr No, there's no "vector" virtual class, so implementation would be a bit tedious. But we don't need to have methods for Date and factor etc, because those will be caught by the method for the underlying atomic vector.

It seems like we're adding new functionality that previously was an error, so it doesn't seem too dangerous to me.

[krlmlr]

@jennybc: For now you could try kimisc::list_to_df() -- I totally forgot about this guy. I still think this should be part of tibble.

[ijlyttle]

Apologies if this is not helpful, but could purrr::map_df be useful?

library("dplyr")
library("purrr")

x <- list(alpha = 'horrible', beta = 'list', gamma = 'column')

x %>% map_df(~ data_frame(thing = .x), .id = "name")

Could a new verb be put in place of the function within map_df?

[krlmlr]

Note that this is already possible with #71:

x %>% as_data_frame %>% rownames_to_column

[github-actions]

This old thread has been automatically locked. If you think you have found something related to this, please open a new issue and link to this old issue if necessary.